[RASMB] precision limit in quantifying protein aggregates by sedimentation velocity with Sedfit

[Peter Schuck]

Yi-Ming and RASMB,

with a little delay I'd like to make another comment on the question of the 
precision for detecting aggregates.  I should say I don't have a lot of 
practical experience in this question myself, like John and Arthur have, 
because this is not the type of problem we're usually studying.  However, I 
think it is useful to do simulations to find out what's theoretically 
possible, an issue separate from the question what experimental limitations 
might be.

For example, assuming a signal of 1 with a noise of 0.005, a 10 mm column, 
and a rotor speed of 50,000 rpm with 50 scans at 150 sec intervals, I find 
that a trimer at 0.2% should be detectable.   That was in a fit floating TI 
and RI noise and the meniscus (starting off with an error by 0.01 in the 
meniscus and optimizing it with the Simplex routine).  The returned value 
was 0.19%, with errors limits from Monte-Carlo analysis of 0.17% to 
0.20%.  That seems to support what John found experimentally.

Another perhaps relevant comment is regarding the data analysis strategy - 
beyond just looking at the c(s) peaks, there are two other approaches that 
I would favor.  One would be to look at the integral of the distribution 
over a certain range, rather than the location and area of a particular 
peak, which can be more variable.   The other option is to use SEDPHAT in 
the hybrid discrete/continuous model, and to treat the monomer and a few 
oligomers as discrete species, constraining the molar masses to be in 
multiples.

Best,
Peter Schuck